On the Relation Between Yangians, Affine Hecke Algebras and Long-Range Interacting Models
Total Page:16
File Type:pdf, Size:1020Kb
Recherche Coopérative sur Programme no 25 DENIS BERNARD On the Relation Between Yangians, Affine Hecke Algebras and Long-Range Interacting Models Les rencontres physiciens-mathématiciens de Strasbourg - RCP25, 1995, tome 47 « Conférences de M. Audin, D. Bernard, A. Bilal, B. Enriquez, E. Frenkel, F. Golse, M. Katz, R. Lawrence, O. Mathieu, P. Von Moerbeke, V. Ovsienko, N. Reshetikhin, S. Thei- sen », , exp. no 3, p. 77-91 <http://www.numdam.org/item?id=RCP25_1995__47__77_0> © Université Louis Pasteur (Strasbourg), 1995, tous droits réservés. L’accès aux archives de la série « Recherche Coopérative sur Programme no 25 » implique l’ac- cord avec les conditions générales d’utilisation (http://www.numdam.org/conditions). Toute utili- sation commerciale ou impression systématique est constitutive d’une infraction pénale. Toute copie ou impression de ce fichier doit contenir la présente mention de copyright. Article numérisé dans le cadre du programme Numérisation de documents anciens mathématiques http://www.numdam.org/ On the relation between Yangians, affine Hecke algebras and long-range interacting models. Denis Bernard 2 Service de Physique Théorique de Saclay 3 F-91191, Gif-sur-Yvette, France. Plan : 1- Long-Range Interacting Models. 2- Yangians and Affine Hecke Algebras. 3- A Yangian deformation of the W-algebras. 1 Long-Range Interacting Models. There is a large family of integrable long range interacting spin chains. In particular, a model introduced by Haldane and Shastry [2], see also [3], is a variant of the spin half Heisenberg chain, with exchange inversely proportional to the square distance between the spins. It possesses the remarkable properties that its spectrum is highly degenerate and additive, and that the elementary excitations are spin half objects obeying a half-fractional statistics intermediate between bosons and fermions. They are defined as follows. We consider a spin chain with TV sites, labeled by integers z, j, · · · ranging from 1 to N. On each sites there is a spin variable σζ· which takes two values: G{ — ±. The hamiltonians, which are all su(2) invariant, are of the following form : Η = Σ h» (Ρ»-1) (1) where P{j is the operator which exchanges the spins at the sites i and j. For translation invariance h{j = h(i — j). Demanding the integrability of the model selects the functions h. 2Member of the CNRS 3Laboratoire de la Direction des Sciences de la Matière du Commisariat à l'Energie Atomique. 77 The possible choices are : {ύή£Ίχ)2, hyperbolic model (7 real) π h(x) = < / ^2, trigonometric model k Pf^), elliptic model. where V(x) is the Weierstrass function. When 7 —• 00, the hyperbolic model reduces to the Heisenberg spin chain: h{j = #t,j+i + and for 7 —» 0, the interaction becomes the 1/x2 exchange. The hyperbolic model has not been completly solved for general 7, although a partial list of eigenstates is known. The elliptic model is even more intriguing since it interpolates between the Heisenberg spin chain of finite length and the trigonometric model [4]· The Haldane-Shastry spin chain is the trigonometric model. In the thermodynamical limit, Ν —> oo, it reduces to the l/x2 exchange model, but it also possesses remarkable properties at finite TV. Notably, its hamiltonian commutes with an infinite dimensional algebra whose two first generators are [5] : Qo = Σ & (2) i Qi = Σ cot9 (^^Tr) & x $ (3) with Si the spin operators acting on the site i. The first generators are the usual su(2) generators. Together with the second ones, they form a representation of the su(2) Yangian, (which is a deformation of the su(2) current algebra, see section 3 for an introduction to the Yangians). This infinite dimensional symmetry is at the origin of the large degeneracy of the spectrum. The fact that the hamiltonian is Yangian invariant at finite ΛΓ is particular to the Haldane-Shastry spin chain; in the Heisenberg spin chain, the Yangian symmetry only appears in the thermodynamical limit. In order to grasp the rules describing the spectrum, we first construct few eigenstates. Clearly, the ferromagnetic vacuum |Ω) = | + Η + +) is an eigenstate : its energy is zero. The eigenstates in the one-magnon sector are the plane waves :\k) = Ση exp(i2Kkn/N)a~ |Ω), with pseudo-momentun fc, 1 < k < (Ν — 1): the one-magnon energy is e(k) — k(k — TV). In the two-magnon sectors, i.e. for states of the form \φ) = Ση,™ ^η,™0"~σ~ |Ω), the eigenstates which are not degenerate with the zero or one-magnon eigenstates are labeled by two pseudo-momenta &i, &2, with 1 < fci, k2 < (Ν — 1). They are given by : , ,71 I . ,771 ^[^1,^2] /L _ L\ (UJnki-]rmk2 . (jmk1^nk2\ ' fLJnk1+mk2 _ UJmk1+nk2\ with ω = exp(i2w/N). Note that these wave functions vanish if ki = k2 but also if \ki — k2\ = 1. The energy of \φ^Μ) is Ε = c(ifci) + e(k2). From the two-magnon computation we learn two properties of the spectrum : (i) it is additive, e.g. the two-magnon energy is the sum of the one-magnon energies, but (ii) the pseudo-momenta satisfy a selection rule : they are neither equal nor they differ by a unit. These rules are the general rules, and the full spectrum can be described as follows [6]. 78 To each eigenstate multiplet is associated a set of pseudo-momenta {kp} which are non- consecutive integers ranging from 1 to (TV — 1). The energy of an eigenstate |{&p}) with pseudo-momenta {kp} is: ' 7Γ N 2 H\{k }) = e(kp) \{kp}} with e(k) = k{k - TV) p (ς (4) Furthermore, the degeneracy of the multiplet with pseudo-momenta {kp} is described by its su(2) representation content as follows. Encode the pseudo-momenta in a sequence of (TV — 1) labels 0 or 1 in which the l's indicate the positions of the pseudo-momenta; add two O's at both extremities of the sequence which now has length (TV + 1). Since the pseudo- momenta are neither equal nor consecutive, two labels 1 cannot be adjacent. The sequence corresponding to the ferromagnetic vacuum is a line of 0, those of the one-magnon states have TV label 0 and only one label 1, and so on. A sequence can be decomposed into the product of elementary motifs, which are series of (Q + 1) consécutives O's. The multiplicity of the spectrum is recovered if to each elementray motif of length (Q + 1) we associate a spin Q/2 representation of su(2). The representation content of the full sequence is then given by the tensor product of its elementary motifs. The magnons are the excitations over the ferromagnetic vacuum; the excitations over the antiferromagnetic vacuum are conveniently described in terms of spinons. For Ν even, the antiferromagnetic vacuum corresponds to the alternating sequence of symbols 010101 · · · 010. The excitations are obtained by flipping and moving the symbols 0 and 1. Let us give the sequences corresponding to the first few excitations over the antiferromagnetic vacuum, (for concreteness we choose TV = 10) : ' 01010101010, antiferromagnetic vacuum (o) < 0 10 10 1 O^O^O 10, a two-spinon excitations (2a) w 0 1 0^0 1 0^0 10 10, a two-spinon excitations (2b), etc... We have inserted a χ between any two consecutive labels 0. These crosses represent the spinon excitations, their number is the spinon number. Note that there is no one-spinon excitation for TV even. By convention, we will say that consecutive crosses not separated by any label 1 correspond to spinons in the same orbital, while crosses separated by labels 1 correspond to spinons in different orbitals. From the rules described above, it follows that the degeneracy of the excitations (2a) and (2b) are different : it is three in the case (2a) and four in the case (2b). These degeneracy are recovered by giving a su(2) spin half to the spinons and by assuming that spinons in the same orbital are in a fully symmetric states. Hence, in the case (2a), there are two spinons in the same orbital and therefore they form a spin one representation of su(2), and in the case (2b), the two spinons are in two different orbitals and therefore they form a su(2) representation isomorphic to the tensor product of two spin half representations of su(2). The fact that the spinons are spin half excitations can also be seen by looking at the excitations of a spin chain of length TV with TV odd. This description of the states generalizes to the full spectrum. We can classify the se quences by their number M of pseudo-momenta. The spinon number Nsp of a sequence is sp then defined by M = ~2 . Since M is an integer, (N—Nsp) is always even : this means that the spinons are always created by pairs. A sequence of pseudo-momenta {kp; ρ =!,···, M}, 79 in the Nsp spinon sector, can be decomposed into (M +1) elementary motifs where, as before an elementary motif is a series of consecutive 0. We call the elementary motifs the accessible N sp orbitals to the spinons. At fixed Nsp, there are Norb = (l + ~^ ) orbitals. Hence, a sequence of pseudo-momenta {kp} corresponds to the filling of the Nori orbitals with respective spinon occupation numbers np = (kP+i — kp — 2), with k0 — — 1 and — Ν + 1 by convention. Since an elementary motif of length (Q + 1) corresponds to a spin Q/2 representation of su(2), the full degeneracy of the sequences is then recovered by assuming that the spinons are spin half objects which behave as bosons in each orbitals.